def save_hdf5(filename, obj):
gpu = (hasattr(obj, "xp") and obj.xp == cuda.cupy)
if gpu: obj.to_cpu()
serializers.save_hdf5(filename, obj)
if gpu: obj.to_gpu()
python类save_hdf5()的实例源码
def save(self):
serializers.save_hdf5("conv.model", self.conv)
if self.fcl_eliminated is False:
serializers.save_hdf5("fc.model", self.fc)
def save_model(self, model_filename):
"""Save a network model to a file
"""
serializers.save_hdf5(model_filename, self.model)
serializers.save_hdf5(model_filename + '.opt', self.optimizer)
def save(self, filename):
tmp_filename = str(uuid.uuid4())
serializers.save_hdf5(tmp_filename, self)
if os.path.isfile(filename):
os.remove(filename)
os.rename(tmp_filename, filename)
def save(self, filename):
tmp_filename = str(uuid.uuid4())
serializers.save_hdf5(tmp_filename, self)
if os.path.isfile(filename):
os.remove(filename)
os.rename(tmp_filename, filename)
def save_model(dirname, model):
model_filename = dirname + "/model.hdf5"
param_filename = dirname + "/params.json"
try:
os.mkdir(dirname)
except:
pass
if os.path.isfile(model_filename):
os.remove(model_filename)
serializers.save_hdf5(model_filename, model)
params = {
"vocab_size_enc": model.vocab_size_enc,
"vocab_size_dec": model.vocab_size_dec,
"ndim_embedding": model.ndim_embedding,
"ndim_h": model.ndim_h,
"num_layers": model.num_layers,
"densely_connected": model.densely_connected,
"pooling": model.pooling,
"zoneout": model.zoneout,
"dropout": model.dropout,
"weightnorm": model.weightnorm,
"wgain": model.wgain,
"attention": isinstance(model, AttentiveSeq2SeqModel),
}
with open(param_filename, "w") as f:
json.dump(params, f, indent=4, sort_keys=True, separators=(',', ': '))
def save_model(dirname, qrnn):
model_filename = dirname + "/model.hdf5"
param_filename = dirname + "/params.json"
try:
os.mkdir(dirname)
except:
pass
if os.path.isfile(model_filename):
os.remove(model_filename)
serializers.save_hdf5(model_filename, qrnn)
params = {
"vocab_size": qrnn.vocab_size,
"ndim_embedding": qrnn.ndim_embedding,
"ndim_h": qrnn.ndim_h,
"num_layers": qrnn.num_layers,
"kernel_size": qrnn.kernel_size,
"pooling": qrnn.pooling,
"zoneout": qrnn.zoneout,
"dropout": qrnn.dropout,
"weightnorm": qrnn.weightnorm,
"wgain": qrnn.wgain,
"densely_connected": qrnn.densely_connected,
"ignore_label": qrnn.ignore_label,
}
with open(param_filename, "w") as f:
json.dump(params, f, indent=4, sort_keys=True, separators=(',', ': '))
def save(self):
serializers.save_hdf5("fc.model", self.fc)
print "model saved."
serializers.save_hdf5("fc.optimizer", self.optimizer_fc)
print "optimizer saved."
def save(self):
serializers.save_hdf5("fc_value.model", self.fc_value)
serializers.save_hdf5("fc_advantage.model", self.fc_advantage)
print "model saved."
serializers.save_hdf5("fc_value.optimizer", self.optimizer_fc_value)
serializers.save_hdf5("fc_advantage.optimizer", self.optimizer_fc_advantage)
print "optimizer saved."
def save(self, folder, epoch, batch):
print('-'*5 , 'saving model')
serializers.save_hdf5('{}/network_epoch{}_batch{}.model'.format(folder, epoch, batch), self.Networks[0])
print('-'*5 , 'saving optimizer')
serializers.save_hdf5('{}/network_epoch{}_batch{}.state'.format(folder, epoch, batch), self.Optimizer)
return
def save(self, model_filename, optimizer_filename):
""" Save the state of the model & optimizer to disk """
serializers.save_hdf5(model_filename, self.model)
serializers.save_hdf5(optimizer_filename, self.optimizer)
def save_params(self, epoch):
print "==> saving state %s" % self.out_model_dir
serializers.save_hdf5("%s/net_model_classifier_%d.h5"%(self.out_model_dir, epoch),self.network)
def save_params(self, epoch):
print "==> saving state %s" % self.out_model_dir
serializers.save_hdf5("%s/net_model_enc_%d.h5"%(self.out_model_dir, epoch),self.enc)
serializers.save_hdf5("%s/net_model_dec_%d.h5"%(self.out_model_dir, epoch),self.dec)
def save(self, filename):
if os.path.isfile(filename):
os.remove(filename)
serializers.save_hdf5(filename, self)
def save(self, filename):
tmp_filename = filename + "." + str(uuid.uuid4())
serializers.save_hdf5(tmp_filename, self)
if os.path.isfile(filename):
os.remove(filename)
os.rename(tmp_filename, filename)
def saveModels(self, savemodelName):
print('save the model')
serializers.save_hdf5(savemodelName, self.model)
print('save the optimizer')
serializers.save_hdf5(savemodelName[:-5]+ 'state', self.optimizer)
def save(self,filename):
cs.save_hdf5(filename,self.func.to_cpu())
def save(self,filename):
#cs.save_hdf5(filename,self.func.copy().to_cpu())
cs.save_hdf5(filename,self.func.copy())
def save(self,filename):
cs.save_hdf5(filename,self.func.to_cpu())
def save(self,filename):
cs.save_hdf5(filename,self.model.copy().to_cpu())
def save(self,filename):
cs.save_hdf5(filename,self.func.copy().to_cpu())
def save(self):
serializers.save_hdf5("conv.model", self.conv)
serializers.save_hdf5("fc_value.model", self.fc_value)
serializers.save_hdf5("fc_advantage.model", self.fc_advantage)
def save_model(dirname, model):
model_filename = dirname + "/model.hdf5"
param_filename = dirname + "/params.json"
try:
os.mkdir(dirname)
except:
pass
if os.path.isfile(model_filename):
os.remove(model_filename)
serializers.save_hdf5(model_filename, model)
params = {
"vocab_size": model.vocab_size,
"ndim_embedding": model.ndim_embedding,
"ndim_h": model.ndim_h,
"num_layers_per_block": model.num_layers_per_block,
"num_blocks": model.num_blocks,
"kernel_size": model.kernel_size,
"dropout": model.dropout,
"weightnorm": model.weightnorm,
"wgain": model.wgain,
"ignore_label": model.ignore_label,
}
with open(param_filename, "w") as f:
json.dump(params, f, indent=4, sort_keys=True, separators=(',', ': '))
def save(self, filename):
if os.path.isfile(filename):
os.remove(filename)
serializers.save_hdf5(filename, self)
def save(self, model_dir="./"):
try:
os.mkdir(model_dir)
except:
pass
serializers.save_hdf5(model_dir + "/wavenet.model", self.chain)
serializers.save_hdf5(model_dir + "/wavenet.opt", self.optimizer)
def save(self, filename):
if os.path.isfile(filename):
os.remove(filename)
serializers.save_hdf5(filename, self)
def make_snapshot(self, model):
# TODO: get model from Optimizer
prefix = "{}_{}".format(self.out_file, self.optimizer.epoch)
serializers.save_hdf5(prefix + ".model", model)
serializers.save_hdf5(prefix + ".state", self.optimizer)
print("Snapshot created")
def save(self):
dir = "model"
serializers.save_hdf5(dir + "/dqn_fc.model", self.fc)
print "model saved."
serializers.save_hdf5(dir + "/dqn_fc.optimizer", self.optimizer_fc)
print "optimizer saved."
def save(self):
dir = "model"
serializers.save_hdf5(dir + "/bddqn_shared_fc.model", self.shared_fc)
serializers.save_hdf5(dir + "/bddqn_shared_fc.optimizer", self.optimizer_shared_fc)
serializers.save_hdf5(dir + "/bddqn_head_fc.model", self.head_fc)
serializers.save_hdf5(dir + "/bddqn_head_fc.optimizer", self.optimizer_head_fc)
def save(self, filename):
if os.path.isfile(filename):
os.remove(filename)
serializers.save_hdf5(filename, self)
